PUBLICATIONS
Geochemistry, Geophysics, Geosystems
RESEARCH ARTICLE
10.1002/2014GC005394
Key Points:
3-D effects on the thermal structure
of the subduction zone are
investigated
The effects on slab surface
temperature are moderate
The effects on surface heat flow are
insignificant
Correspondence to:
M. Morishige,
mmorishige@jamstec.go.jp
Citation:
Morishige, M., and P. E. van Keken
(2014), Along-arc variation in the 3-D
thermal structure around the junction
between the Japan and Kurile arcs,
Geochem. Geophys. Geosyst., 15, 2225–
2240, doi:10.1002/2014GC005394.
Received 23 APR 2014
Accepted 10 MAY 2014
Accepted article online 15 MAY 2014
Published online 4 JUN 2014
Along-arc variation in the 3-D thermal structure around the
junction between the Japan and Kurile arcs
Manabu Morishige1 and Peter E. van Keken2
1
Department of Deep Earth Structure and Dynamics Research, Japan Agency for Marine-Earth Science and Technology,
Yokosuka, Japan, 2Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, Michigan, USA
Abstract The thermal structure in subduction zones has a strong influence on seismogenesis and arc
volcanism. Traditional 2-D models have been used to provide reasonable agreement between models
and observations, but in a number of cases clear 3-D effects are present. One such case is in the Northern Japan subduction system. At the junction between Japan and Kurile arcs, surface heat flow and the
occurrence of intermediate-depth seismicity are different than in the Tohoku and Hokkaido regions. We
investigate the effects of 3-D slab geometry and a local deepening of slab-mantle decoupling depth on
the thermal structure in this region based on 3-D finite element approach. We find that both effects produce the along-arc variation of slab surface temperature, which could reach 100 C. The warmer region
arises through 3-D effects of thermal conduction and the colder region arises through localized slow
incoming flow in the case where 3-D slab geometry is taken into account. 3-D flow arises where a local
deepening of slab-mantle decoupling depth is assumed, which leads to both warmer and colder regions.
The effects on surface heat flow are small. While intermediate-depth seismicity in the subducted crust is
suggested to be controlled by temperature-dependent phase transitions, the predicted changes in thermal structure are not sufficient to cause the observed deepening of seismicity. This suggests that the
thermal structure of this subduction zone may be more strongly influenced by time-dependent deformation of the overriding crust and slab.
1. Introduction
Our understanding about the thermal structure of the subduction zone has improved in the last few decades. The main factors controlling the thermal structure of the subduction zone are the heat advection due
to cold subducting slab and hot flowing mantle. Thermal conduction also plays an important role near the
slab surface and in the overriding plate.
Of particular importance is the cornerflow which arises due to the coupling between downgoing slab and
overriding mantle wedge. The rheology of the wedge itself critically influences the heat transport to the
slab, and therefore, directly controls the slab surface temperatures. van Keken et al. [2002], Kelemen et al.
[2003], and Conder [2005] demonstrated that introducing a realistic olivine rheology leads to the focus of
the flow in the mantle wedge. As a result, the temperature increases in the mantle wedge and on the slab
surface by a few hundred degrees compared to older isoviscous models. The high below-arc temperatures
are confirmed by a detailed benchmark for subduction zone models suggesting the governing equations
are solved with high accuracy [van Keken et al., 2008]. The effect of radiogenic heat production is small and
primarily affects the thermal structure in the overriding plate [Hall, 2012]. Constraints from geophysical
observations, in particular those of heat flow and seismic tomography [Tsumura et al., 2000; Bostock et al.,
2002; Hyndman and Peacock, 2003; Schurr et al., 2003; Tanaka et al., 2004; Eberhart-Phillips et al., 2008; Rondenay et al., 2008], provide an indication that the mantle wedge in the fore arc is anomalously cold and apparently not involved in the cornerflow that is induced by the subducting slab. We can assume that over
geological timescales the slab is effectively decoupled from the overlying plate due to frequent underthrusting earthquakes along the seismogenic zone that generally ends near 50 km depth. The observed
cold corner requires that the slab is decoupled to greater depth by creeping or other aseismic but local
deformation. In northern Japan, as well as in other well-instrumented subduction zones, the geometry of
the cold corner is best explained when the slab couples to the overlying mantle wedge at depth of approximately 80 km [Wada and Wang, 2009].
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2225
Geochemistry, Geophysics, Geosystems
46˚
25
0
45˚
20
0
15
44˚
250
0
10
Hokkaido
200
0
50
150
43˚
Ku
50
150
41˚
100
42˚
rile
e
Tr
nc
39˚
Japan Trenc
Tohoku
h
40˚
h
10.1002/2014GC005394
Honda [1985] first proposed using
numerical model that the rigid fore-arc
mantle is necessary to explain the low
surface heat flow region in the northeast Japan. Wada and Wang [2009]
extended the work to a wide range of
subduction zones and proposed the
idea of common depth of slab-mantle
decoupling. The slab-mantle decoupling depth is a depth such that the
mantle does not move with the slab
above it and the mantle moves with
the same velocity as the slab beneath
it. The transition in the rheology and
resulting nonlinear feedback between
thermal structure and flow geometry
causes the near-stagnation of the shallow wedge with resulting lower temperatures [Kneller et al., 2007; Wada
et al., 2011].
The use of thermal modeling is particularly useful in direct comparisons with
the temperature estimates from arc
geochemistry and petrology. In recent
37˚
83.5 mm/yr (PA−NA)
years, we have seen good agreement
Kanto
between model predictions and geoheatflow (mW/m )
36˚
chemical observations. Application of
100 km
50
75 100 125
the H2 O-Ce thermometer to determine
slab fluid temperatures has shown
35˚
139˚ 140˚ 141˚ 142˚ 143˚ 144˚ 145˚ 146˚ 147˚ 148˚
good correspondence with a specific
thermal model for Kamchatka [Plank
Figure 1. Summary of observations near the junction between Japan and Kurile
et al., 2009]. Application of the same
arcs based on Figure 1 of van Keken et al. [2012]. Thick black line shows the locatechnique to 10 subduction zones
tion of Japan and Kurile trench. Thin black lines are the contours of slab surface
every 10 km [Zhao et al., 1997; Nakajima and Hasegawa, 2006; Kita et al., 2010].
[Cooper et al., 2012] also show remarkColored circles show surface heat flow [Tanaka et al., 2004]. Red line shows the
able agreement between geochemical
downdip limit of low-angle thrust events [Kita et al., 2010]. Blue line shows the
estimates of the slab surface temperalocation where we show the cross section in Figure 8b. Arrow shows relative
movement of Pacific plate (PA) to North America plate (NA).
ture to those predicted in a global
study using two-dimensional finite element models [Syracuse et al., 2010].
The models from Syracuse et al. [2010] also compare favorably with trace and major element mixing models
for the Chilean [Jacques et al., 2013; Watt et al., 2013] and the Philippine arcs [Turner et al., 2011]. Since geochemical and petrological analyses can be diagnostic for temperature variations that are significantly
smaller than 100 C it is critical for us to develop an understanding of how 3-D variations in subduction
geometry and flow can influence slab and wedge temperatures.
50
100
150
38˚
2
In this study, we consider the region around the junction between Japan and Kurile arcs (Figure 1). The subduction occurs nearly normal to the trench beneath Japan arc (Tohoku region), whereas an oblique subduction occurs beneath Kurile arc (Hokkaido region). Numerous studies on the Tohoku thermal structure have
been made with 2-D modeling [e.g., Honda, 1985; Peacock and Wang, 1999; van Keken et al., 2002, 2012],
which appears appropriate due to limited along-arc variations. There are only a few studies focusing on an
oblique subduction with 3-D models which is appropriate for the transition into Hokkaido [e.g., Honda and
Yoshida, 2005; Kneller and van Keken, 2008; Bengtson and van Keken, 2012]. The number of numerical studies
which consider these two regions together is quite small [Morishige et al., 2010; Morishige and Honda, 2013].
Morishige and Honda [2013] showed that the along-arc variation of seismic anisotropy [e.g., Nakajima et al.,
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2226
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
2006], subduction angle [e.g., Hayes et al., 2012], and the slab behavior in the transition zone [e.g., Fukao
et al., 2001; Miller and Kennett, 2006] can be explained at least qualitatively by the effects of the trench
shape.
In this study, we focus on two other types of observations in this region which are affected by the thermal
structure of the subduction zone.
The surface heat flow observations for northern Japan [Tanaka et al., 2004] show the sharp transition from
low values in the fore arc to high and scattered values in the arc and back arc. In southern Hokkaido, an
anomalous incursion of low heat flow values occurs toward the Japan Sea (Figure 1). This suggests that the
crust and mantle beneath this region is anomalously cold. Kita et al. [2010, 2012] imaged an anomalously
low seismic velocity area below this region and proposed the presence of subducted continental crust. It
may be caused by the southwestward migration of Kurile fore-arc sliver due to oblique subduction of Pacific
plate since the middle Miocene [Kimura, 1986]. The presence of this block was also suggested to cause the
anomalous deepening of the seismic belt in this region [Kita et al., 2010] although a subsequent study that
took this lower crustal root into account could not conclusively link the shift in seismicity to the modified
thermal structure [van Keken et al., 2012].
The upper plane of intermediate-depth seismicity below southern Hokkaido is different from its surroundings. While the seismicity of the upper plane remains in the crust, it occurs deeper than in the surrounding
regions to the north and south [Kita et al., 2010] and it gets closer to the slab surface in deeper part,
whereas the seismicity diverges from slab surface with depth in other regions [van Keken et al., 2012]. It suggests a modified dynamic or thermal regime there [Kita et al., 2010]. van Keken et al. [2012] showed with a
series of 2-D numerical models that the location of the blueschist-eclogite transition in the subducted oceanic crust overlaps with the termination of crustal seismicity in Tohoku and Hokkaido, but not beneath the
junction between them.
We note that an anomalous area with deepened seismicity and an extended low surface heat flow region
occurs also beneath Kanto [Tanaka et al., 2004; Hasegawa et al., 2007], but this is beyond the scope of this study.
In this study, we will test whether the anomalous heat flow and intermediate-depth seismicity are caused
simply by 3-D effects near this junction, as was proposed in van Keken et al. [2012]. Toward this goal, we
consider three types of cases. The first case focuses on the effects only of 3-D slab geometry. In the second
case, we take into account the possible deepening of the transition from underthrusting and slow slip
events to full coupling along the plate interface. The downdip limit of low angle thrust earthquakes beneath
the junction is deeper than its surroundings by around 15 km (red line in Figure 1), which may be possibly
due to subducted fore-arc crust [Kita et al., 2010]. In the third case, we consider the two types of effects
mentioned above together.
2. Modeling Approach
We divide the model domains into three parts: the crust, the mantle wedge, and the subducting slab. The
x-z cross section is the same along the strike of the trench (y-direction) in all the cases considered except
for the slab-mantle decoupling depth as we will show later (Figure 2a). The crust thickness is 25 km and the
subduction angle is 30 . We consider three cases where slab geometry and/or the slab-mantle decoupling
depth differ (Figure 2c).
In Case 1, we use the slab geometry similar to that of the junction between Japan and Kurile arcs to see the
effects of 3-D slab geometry on the thermal structure. Slab-mantle decoupling depth (Ddec ) is constant
along the strike of the trench and set to be 80 km [Wada and Wang, 2009].
In Case 2, we use a simple slab geometry but include a local deepening of slab-mantle decoupling depth
which is constrained from the along-arc variation of downdip limit of low angle thrust earthquakes [Kita
et al., 2010]. It is given as Ddec ½km 587:517:5sin ððy2175Þp=50Þ (at y ½km 150) and 80 (at y [km] < 150),
which means that Ddec ranges from 80 to 95 km.
In Case 3, we use the same slab geometry as that used in Case 1, but the slab-mantle decoupling depth is
given as Ddec ½km 587:517:5sin ððy2275Þp=50Þ (at 250 km y 350 km) and is kept constant at 80 km
elsewhere.
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2227
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Figure 2. Numerical models. (a) Model geometry and boundary conditions in x-z cross section. (b) An example of mesh in the region enclosed by a dashed line in Figure 2a. (c) Model
geometries and boundary conditions in 3-D view. The green points in Figures 2a and 2b, and the green lines in Figure 2c show the location of slab-mantle decoupling depth.
The mantle flow is computed only in the mantle wedge by solving the equations of mass conservation and
motion for an incompressible infinite Prandtl number fluid without buoyancy forces:
r v50;
(1)
r s 2rp50;
(2)
s 52g_ ;
(3)
where v is velocity, s is the deviatoric stress tensor, p is dynamic pressure, g is effective viscosity, and _ is
strain rate tensor defined by
_ 5
1 T rv 1 rv
:
2
(4)
We assume a rigid boundary v 50 between the mantle wedge and the crust. The velocity boundary condition at the slab surface depends on the depth considered. Below the slab-mantle decoupling depth, we
impose v 5v slab 5ðV0 cos 30o ; 0; 2V0 sin 30o Þ with V0 58 cm/yr while above this depth the velocity
boundary condition is zero. v changes from 0 to v slab linearly over 6 km along the slab surface around the
slab-mantle decoupling depth to mitigate the effects of the pressure singularity that arises with a discontinuous velocity boundary condition as discussed in van Keken et al. [2002]. The wedge inflow/outflow boundaries are assumed to be 2p1sxx 52p1syy 52p1szz 50 leading to free inflow and outflow. We assume
vy 5 0 cm/yr and zero shear stress at the boundaries in y-direction.
We solve the time-dependent heat equation for the whole domain. We ignore radiogenic heat production
and assume the Boussinesq approximation:
@T
1v rT5jr2 T;
@t
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
(5)
2228
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
where T is temperature, t is time, and j (57.272 3 1027 m2/s) is thermal diffusivity. In solving this equation,
the velocity in the crust and subducting slab are assumed to be 0 and v slab , respectively. The surface
boundary condition is 0 C and the GDH1 model [Stein and Stein, 1992] is used to define the temperature at
the wedge inflow (assuming an age of 30 Myr appropriate for the Japan Sea) and slab inflow (with an age
of 130 Myr). GDH1 model is a plate cooling model where the plate thickness is 95 km and the potential temperature at the bottom of the plate is 1421.5 C. No heat flow is assumed at other boundaries. We assume
the constant adiabatic temperature gradient of 0.3 K/km that we add as a postprocessing step.
We consider dislocation creep in the mantle wedge. The effective viscosity for this deformation mechanism
can be written as
gdisl ðT; _ Þ5Adisl exp
Edisl ð12nÞ=n
_
;
nRT
(6)
where Adisl 52:8968 3 104 Pa s1=n is constant, Edisl 5540 kJ/mol is activation energy, R 5 8.3145 J/K/mol is
the gas constant, n 5 3.5 is a constant, and _ is the second invariant of strain rate tensor defined as
rffiffiffiffiffiffiffiffiffiffiffiffi
1
_ ij _ ij :
_ 5
2
(7)
The values of these parameters are based on Karato and Wu [1993] for dry olivine.
The effective viscosity used in equation (3) is defined by
1
1 21
g5
1
;
gdisl gmax
(8)
where gmax 51024 Pa s is the maximum viscosity.
We use the finite element code Sepran [Cuvelier et al., 1986] with a linear tetrahedral mesh to solve these
equations. The minimum linear element size is 1.5 km around the slab-mantle decoupling depth (Figure 2b)
with the resolution of elements increasing to 20 km in areas of small temperature variations. We conducted
convergence tests for an essentially 2-D case (3-D case with no along-arc variation) and found that this resolution is acceptable. We use GDH1 model [Stein and Stein, 1992] with an age of 30 Myr as the initial thermal
structure. We evolve the time-dependent models for 20 Myr and use a time step of 0.38 Myr.
3. Results
In this section, we mainly focus on the slab surface temperature and surface heat flow in each of the three
cases. The surface heat flow is defined by
F52k
dT
;
dz
(9)
where k 5 3 W/m/K is the thermal conductivity.
3.1. Case 1: Effects of 3-D Slab Geometry
Figure 3 shows the results of the case focusing on the effects of 3-D slab geometry on the thermal structure
of the subduction zone. The curvature radius RC of the junction is 141.4 km in this case.
We can see first that the slab surface temperature is slightly higher where an oblique subduction occurs
compared to where the subduction occurs normal to the trench (compare, for example, the temperature at
y 5 150 km and y 5 500 km, Figure 3a). Near the junction, we can see a colder region at y 360 km and a
warmer region at y 255 km on the slab surface. The minor departures near y 5 0 km are caused by the
assumed flow boundary condition there and can be ignored.
Surface heat flow shows the general trend that it is low near the trench and high far from the trench, but
there is no significant along-arc variation (Figure 3b). Horizontal velocity of the mantle flow shows that the
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2229
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 1 (large RC)
(b) Surface heat flow
100
100
200
200
y[km]
0
300
400
−250
400
500
700
300
0
y[km]
(a) Slab surface temperature
0
600
500
400
500
300
200
100
0
700
600
500
x[km]
0
400
300
200
100
0
80
90
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
F[mW/m2]
T[oC]
(c) Horizontal velocity at z=−100 km
0
y[km]
100
200
300
400
:5 cm/yr
500
700
600
500
400
300
200
100
0
5
6
x[km]
0
1
2
3
4
v[cm/yr]
Figure 3. Results of Case 1 with a large curvature radius (RC ). (a) Slab surface temperature. Color shows temperature and white lines show the contours of slab surface every 50 km depth.
Arrow shows the direction of plate motion.q
Green
line shows assumed slab-mantle decoupling depth. (b) Surface heat flow. (c) Horizontal velocity at 100 km depth. Arrows show horizonffiffiffiffiffiffiffiffiffiffiffiffiffi
tal velocity and color shows its magnitude vx2 1vy2 .
direction of incoming flow is nearly normal to the trench, but a region where the flow velocity is low arises
near the junction (at y 350 km, Figure 3c).
We also investigate the effects of the smoothness of the junction (Figure 4). In this case, the curvature radius
RC of the junction is smaller (28.3 km). The colder region on the slab surface at y 340 km is similar to
the case with a larger curvature radius, but we can see the warmer region more clearly at y 295 km in this
case (Figure 4a). The surface heat flow and horizontal velocity of the mantle flow are similar to the case
with a larger curvature radius (Figures 4b and 4c).
3.2. Case 2: Effects of a Local Deepening of Slab-Mantle Decoupling Depth
We show the results of the case focusing on the effects of a local deepening of slab-mantle decoupling
depth (Figure 5). The increase of slab surface temperature at the slab-mantle decoupling depth is larger
where we assume a local deepening of slab-mantle decoupling depth, which produces a warmer region
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2230
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 1 (small RC)
(b) Surface heat flow
100
100
200
200
y[km]
0
300
400
−250
400
500
700
300
0
y[km]
(a) Slab surface temperature
0
600
500
400
500
300
200
100
0
700
600
500
x[km]
0
400
300
200
100
0
80
90
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
F[mW/m2]
T[oC]
(c) Horizontal velocity at z=−100 km
0
y[km]
100
200
300
400
:5 cm/yr
500
700
600
500
400
300
200
100
0
5
6
x[km]
0
1
2
3
4
v[cm/yr]
Figure 4. (a–c) Results of Case 1 with a small curvature radius (RC ) (same as in Figure 3). The arrows on the right side of Figure 4a show the position of the warmer and colder regions
where we consider the cross sections in Figure 8.
there (at y 200 km, Figure 5a). There is a region with a slightly lower temperature just next to it at y
160 km. The along-arc variation of surface heat flow is insignificant (Figure 5b). The mantle flow in the
wedge shows a significant 3-D pattern in the region where a local deepening of the slab-mantle decoupling
depth is assumed (Figure 5c). The mantle flow spreads in the direction parallel to the trench (y-direction)
near the slab-mantle decoupling depth at y 200 km. The material first located near the top moves in
y-direction over a long distance (path 3), whereas the material first located far from the top does not move
significantly in the y-direction (paths 1 and 2). The material where the slab-mantle decoupling depth is constant shows a more common 2-D cornerflow pattern (paths 4 and 5).
3.3. Case 3: Combination of Case 1 and Case 2
Figure 6 shows the results of the case where we consider the combined effects of 3-D slab geometry and a
local deepening of slab-mantle decoupling depth. We observe two warmer regions at y 255 km, 300 km
and a colder region at y 355 km (Figure 6a). There again is little change in surface heat flow along the
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2231
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 2
(a) Slab surface temperature
(b) Surface heat flow
y[km]
0
100
100
0
−250
y[km]
0
200
200
400
300
200
100
0
400
300
200
x[km]
0
100
0
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
80
90
F[mW/m2]
T[oC]
(c)Streamlines
0
0
5 4
z[km]
5
3
4
3
2
1
2
1
z[km]
0
x[km]
5
z
z -250
0
y[km]
y
x
2 1 -250
200
43
460 0
5
y[km]
4
3
2 1
200
y
x
Figure 5. (a, b) Results of Case 2 (as in Figure 3). (c) 3-D views of streamlines with a numbering. The green dashed line shows assumed slab-mantle decoupling depth. The color of
streamlines shows z-coordinate along each streamline.
strike of the trench (Figure 6b). Horizontal velocity of the mantle flow is similar to that in Case 1 with a large
curvature radius of the junction (Figure 3c), although we can see a slight difference near the junction (at
y 350 km, Figure 6c).
3.4. Steady State Cases
The three cases discussed above are based on a 20 Myr evolution of the time-dependent model. This is sufficient to bring the slab thermal structure of 2-D models into a quasi steady state. A longer evolution will
cause changes in the thermal structure of the overriding plate, which is relevant for surface heat flow. Hall
[2012] discussed the difference between the results in steady state and time-dependent cases using simple
2-D subduction zone models, similar to van Keken et al. [2008]. We evaluate how the assumption of pure
steady state, where we do not solve the time-dependent equations but assume @T=@t 5 0 in equation (5),
affects the thermal structure in these 3-D models.
Figures 7a and 7b are the results obtained assuming steady state for Case 1 with a large curvature radius of
the junction. Intriguingly, the slab surface temperature clearly differs from that in time-dependent case for
Case 1. Specifically, the colder region (at y 390 km) is more pronounced (cf. Figures 3a and 7a).
On the other hand, the difference in slab surface temperature between steady state and time-dependent
cases for Case 2 is rather small (cf. Figures 5a and 7c).
The surface heat flow shows slightly more distinct along-arc variation in the steady state cases (Figures 7b
and 7d). The thermal structure of the overriding plate is somewhat more affected by the underlying thermal
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2232
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 3
(b) Surface heat flow
100
100
200
200
y[km]
0
300
400
−250
400
500
700
300
0
y[km]
(a) Slab surface temperature
0
600
500
400
500
300
200
100
0
700
600
500
x[km]
0
400
300
200
100
0
80
90
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
F[mW/m2]
T[oC]
(c) Horizontal velocity at z=−100 km
0
y[km]
100
200
300
400
:5 cm/yr
500
700
600
500
400
300
200
100
0
5
6
x[km]
0
1
2
3
4
v[cm/yr]
Figure 6. Results of Case 3 with descriptions as in Figure 4.
structure. The along-arc variation of surface heat flow in Case 2 directly reflects the assumed slab-mantle
decoupling depth. Note that the values of surface heat flow are lower in steady state cases as a whole,
which is caused by the conductive thickening of lithosphere of the overriding plate.
4. Discussion
4.1. Slab Surface Temperature
We showed that the along-arc variation of slab surface temperature arises by the effects of 3-D slab geometry and a local deepening of slab-mantle decoupling depth. In this subsection, we consider the possible
causes of these thermal anomalies. First, we saw that slab surface temperature is generally higher where an
oblique subduction occurs compared to where the subduction occurs normal to the trench. It can be
explained by the effects of subduction obliquity. Bengtson and van Keken [2012] showed that the slab surface temperature in an oblique subduction case was identical to that obtained in a 2-D cross section which
is taken normal to the trench. We consider a proxy of the temperature in the slab called thermal parameter
U [Kirby et al., 1991] (which is defined as U5Avp sin h where A is plate age, vp is slab speed, and h is
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2233
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 1 (large RC, steady state)
(b) Surface heat flow
100
100
200
200
y[km]
0
300
500
700
300
400
−250
400
0
y[km]
(a) Slab surface temperature
0
600
500
400
300
500
200
100
0
700
600
500
x[km]
400
300
200
100
0
x[km]
Case 2 (steady state)
(c) Slab surface temperature
(d) Surface heat flow
y[km]
0
100
100
0
−250
y[km]
0
200
200
400
300
200
100
0
400
300
x[km]
0
200
100
0
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
80
90
2
o
F[mW/m ]
T[ C]
Figure 7. Steady state results. (a, b) Results of Case 1 with a large curvature radius, (c, d) Case 2 where steady state is assumed. Figures 7a and 7c are the same as Figure 3a, and Figures
7b and 7d are the same as Figure 3b.
subduction angle which are taken in the trench-normal direction). In the cross section normal to the trench
U54648 km where an oblique subduction occurs (note that the maximum angle between the direction of
subduction and the strike of the trench is 45 ), but it increases to 5200 km where the subduction occurs
normal to the trench. This increase explains the difference of slab surface temperature in the two regions. In
Case 1, a region with slower incoming flow (Figures 3c and 4c) also causes less rapid warming of the slab
surface near the junction, as is indicated by the overlap in location of slow flow and low slab surface temperature (at y 350 km). Changes in the curvature radius of the junction (RC ) are less important (Figures 3c
and 4c), demonstrating that the large-scale flow in the mantle wedge is not affected by the smoothness of
the junction at least for the range of RC considered in this study.
Interestingly, the warmer region seen in Cases 1 (at y 255 km in Case 1 with large curvature radius and y
295 km in Case 1 with low curvature radius) appear to be caused by thermal conduction as the slab is
warmed from two different horizontal directions (from the left and upper left directions, Figures 3a and 4a),
so the temperature rises. The fact that this effect is weaker when the junction is smoother supports this
idea (Figures 3a and 4a).
In Case 2, we saw that a local deepening of slab-mantle decoupling depth leads to a 3-D flow in the mantle
wedge (Figure 5c). The warmer region at y 200 km arises because the material here originally comes from
relatively deeper and warmer part (paths 1 and 2). The warmer slab surface for a larger decoupling depth is
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2234
Geochemistry, Geophysics, Geosystems
(a1) Blueschist−out boundaries
−80
y=295 km
−100
z[km]
z[km]
(a2)
Case 1 (small Rc)
−80
10.1002/2014GC005394
y=340 km
−120
Case 3
y=300 km
−100
y=355 km
−120
−140
−140
240
220
200
180
160
140
240
x[km]
220
200
180
160
140
x[km]
(b) Cross−section near the junction
(a3)
-40
2D
−80
-60
z[km]
z[km]
-80
−100
−120
-100
-120
-140
−140
240
220
200
180
160
-160
140
-180
x[km]
340 320 300 280 260 240 220 200 180
Distance from trench[km]
(d) Moho temperature
900
800
800
700
700
600
600
500
500
T[oC]
T[oC]
(c) Slab surface temperature
900
400
300
Case 1 (y=295 km)
Case 1 (y=340 km)
Case 3 (y=300 km)
Case 3 (y=355 km)
2D
400
300
Case 1 (y=295 km)
Case 1 (y=340 km)
Case 3 (y=300 km)
Case 3 (y=355 km)
2D
200
100
200
100
0
0
−140 −120 −100
−80
−60
−40
−140 −120 −100
z[km]
−80
−60
−40
z[km]
Figure 8. (a) Predicted blueschist-out boundaries at several cross sections for (a1) Case 1 with small curvature radius, (a2) Case 3, and (a3) 2-D case for comparison. Black lines show
upper and lower boundaries of the oceanic crust. We assume that the width of the oceanic crust is 7 km. Blue lines show predicted blueschist-out boundaries. (b) Cross section near the
junction based on P15 of Figure 6 in van Keken et al. [2012]. The line where the cross section is taken is shown in Figure 1 with blue line. Black line shows slab surface, orange circles are
seismicity within 10 km from the cross section where their size shows magnitude. (c) Slab surface temperature. (d) Moho temperature.
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2235
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
also observed in 2-D models [Wada et al., 2008]. The path 3 includes the motion in y-direction compared to
paths 4 and 5, so it takes longer time for the material to move down to a certain depth on the slab surface
along this path. It means that the material which moves along the path 3 is cooled for longer time, which
leads to the colder region at y 160 km.
The slab surface temperature distribution in Case 3 may be understood simply by combining the results
obtained in Cases 1 and 2. The effects of 3-D slab geometry may cause the warmer region at y 255 km
and the colder region at y 355 km, as we have seen in Case 1. The warmer region at y 300 km is likely
due to 3-D flow caused by a local deepening of slab-mantle decoupling depth as in Case 2.
4.2. Intermediate-Depth Seismicity
We now turn to the question whether the changes in slab temperature can explain the deepening of the
seismic belt below southern Hokkaido as suggested by van Keken et al. [2012] based on observations by
Kita et al. [2010]. Since in most of the Tohoku and Hokkaido subduction systems, the upper plane seismicity
is restricted to the region where blueschist facies rocks are predicted and disappear where eclogite facies
rock should occur [van Keken et al., 2012], we calculate the blueschist-out boundaries for Case 1 (with low
curvature radius) and Case 3 and compare these to a 2-D case (Figures 8a1–8a3). We assume that the
blueschist-out boundary is described by T [ C] 5 617 2 52 P [GPa] where P is pressure [Hacker, 2008]. The
pressure is calculated by using the values 2700 and 3300 kg/m3 as the density of crust and mantle materials,
respectively. First, the result of 2-D case shows that blueschist-out boundary starts at 80 km depth and ends
at 125 km depth. The depth it starts may be controlled mainly by assumed slab-mantle decoupling depth.
The result of Case 1 shows that the depth they start are the same (i.e., at 80 km) for the cross sections in the
cold and warm regions, but the depth where it ends is shallower for the warm region. It is because that the
difference in slab surface temperature remains at Moho depth (red lines in Figures 8c and 8d). In Case 3, on
the other hand, the depth at which the blueschist-out boundaries start differs for the cross sections in the
cold and warm regions, which reflects the change in decoupling depth. However, this difference is recovered deeper in the subducting oceanic crust. It reflects the fact that the difference in Moho temperature
becomes small at deeper part (blue lines in Figure 8d). This feature can be seen also in 2-D numerical models [van Keken et al., 2012]. We also calculate blueschist-out boundaries in the cross sections far from the
junction (at y 5 150 and 500 km) for Cases 1 and 3, and find that the obtained blueschist-out boundaries
start at around 80 km depth and diverge from slab surface with depth, which are broadly consistent with
observed intermediate-depth seismicity below Tohoku and Hokkaido (results not shown).
At the junction between Japan and Kurile arcs, van Keken et al. [2012] showed that the intermediate-depth
seismicity may occur a few kilometer beneath the slab surface in the shallow part and then gradually get
closer to the slab surface with depth (Figure 8b). However, we do not observe this kind of trend at the junction in any of the cases considered in this study. Using a different rheology or a large value of maximum
slab-mantle decoupling depth increases the variation of slab surface temperature but does not help explain
the distribution of the seismicity (Appendix A).
This clearly suggests that the 3-D flow geometry modeled here, even with changes in decoupling depth,
cannot satisfactorily explain the deepening and relative location to slab surface of the seismic belt. One suggestion is that 3-D deformation of the slab (which is ignored here) can be important in causing the changes
in thermal structure and modification of the blueschist-eclogite facies transition.
A second suggestion is that the relative location of the intermediate-depth seismicity in relation to the slab
surface is not as precise beneath the junction as it is further south in Tohoku, since they are not determined
simultaneously there [Kita et al., 2010], although we observe that the correlation between seismicity and
predicted mineral facies is recovered also in Hokkaido north of the junction [van Keken et al., 2012].
An alternative suggestion for the deepening of the seismic belt could also be a kinetic delay of the phase
change, although this begs the question why the kinetic conditions would be different in the junction but
not elsewhere in the Northern Japan subduction zones.
4.3. Surface Heat Flow
In the cases considered in this study, we see only very minor along-arc variation in surface heat flow (Figures 3b, 4b, 5b, and 6b). The variations are slightly larger in the steady state cases (Figures 7b and 7d), but
still much weaker than the clear heat flow excursion that is observed [Tanaka et al., 2004] (Figure 1).
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2236
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 3 (diffusion creep)
(b) Surface heat flow
100
100
200
200
y[km]
0
300
500
700
300
400
−250
400
0
y[km]
(a) Slab surface temperature
0
600
500
400
500
300
200
100
0
700
600
500
400
x[km]
0
300
200
100
0
80
90
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
2
o
F[mW/m ]
T[ C]
(c) Horizontal velocity at z=−100 km
0
(d) Blueschist−out boundary
−80
200
z[km]
y[km]
100
300
y=300 km
−100
y=345 km
−120
400
:5 cm/yr
−140
500
700
600
500
400
300
200
100
0
240
x[km]
0
1
2
220
200
180
160
140
x[km]
3
4
5
6
v[cm/yr]
Figure A1. Results of Case 3 with diffusion creep instead of dislocation creep. Figures A1a–A1c as in Figures 6a–6c. Figure A1d as in Figure 8a.
For now, the best candidate to explain the low surface heat flow region at the junction may be the possible
existence of subducted fore-arc crust [e.g., Kita et al., 2010, 2012].
We plan on future studies that will include the time-dependent behavior of fore-arc crust, a realistic Moho
structure [Katsumata, 2010] and the associated changes in radiogenic heat production, to develop a better
understanding of surface heat flow in this region.
5. Conclusion
We investigated the effects of 3-D slab geometry and a local deepening of slab-mantle decoupling depth
on thermal structure of the subduction zone near the junction between Japan and Kurile arcs. We found
that both effects produced the along-arc variation of slab surface temperature which includes regions that
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2237
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Case 3 (Ddec,max=120 km)
(b) Surface heat flow
100
100
200
200
y[km]
0
300
500
700
300
400
−250
400
0
y[km]
(a) Slab surface temperature
0
600
500
400
500
300
200
100
0
700
600
500
400
x[km]
0
300
200
100
0
80
90
x[km]
100 200 300 400 500 600 700 800 900
0
10
20
30
40
50
60
70
2
o
F[mW/m ]
T[ C]
(c) Horizontal velocity at z=−100 km
0
(d) Blueschist−out boundary
−80
200
z[km]
y[km]
100
300
y=300 km
−100
y=340 km
−120
400
:5 cm/yr
−140
500
700
600
500
400
300
200
100
0
240
x[km]
0
1
2
220
200
180
160
140
x[km]
3
4
5
6
v[cm/yr]
Figure A2. Results of Case 3 with the maximum slab-mantle decoupling depth (Ddec ;max ) extended down to 120 km with descriptions as in Figure A1.
are either warmer or cooler than expected from 2-D models. The warmer region arises through 3-D effects
of thermal conduction and the colder region arises through localized slow incoming flow in the case where
3-D slab geometry is taken into account. 3-D flow arises where a local deepening of slab-mantle decoupling
depth is assumed, which leads to both warmer and colder regions. The along-arc difference of slab surface
temperature could reach 100 C when large maximum slab-mantle decoupling depth (120 km) is assumed
or diffusion creep is considered as the deformation mechanism. Generally, the temperature differences
along the strike of the trench are modest but may be detectable by geochemical methods that determine
independently the temperature of the slab below the arc. Neither the anomalous deepening of the upper
plane seismicity nor the heat flow excursion in southern Hokkaido is explained well by these 3-D models,
suggesting a bigger importance of structural variations in the overriding plate compounded potentially by
3-D dynamical effects that are not taken into account in this study.
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2238
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Appendix A: Effects of Rheology and the Maximum Slab-Mantle Decoupling Depth
We show in this section how the results presented in this paper depends on assumed input parameters.
There may be little uncertainty in 3-D slab geometry and slab velocity. Therefore, we investigate the effects
of changing rheology and the maximum slab-mantle decoupling depth for Case 3.
Figure A1 shows the results of Case 3 where diffusion creep is considered as a deformation mechanism in
the mantle wedge instead of dislocation creep. In this case, the effective viscosity in equation (3) is defined
by
1
1 21
g5
1
;
gdiff gmax
(A1)
where the viscosity for diffusion creep is written as
gdiff ðTÞ5Adiff exp
Ediff
;
RT
(A2)
where Adiff 51:32043109 Pa s is constant and Ediff 5335 kJ/mol is activation energy. The values of these
parameters are based on Karato and Wu [1993] for dry olivine.
Slab surface temperature shows a larger variation compared to the original Case 3 (Figure A1a). It is probably because that the length scale of 3-D flow near y 300 km becomes larger (note that the general
effects of using non-Newtonian viscosity is to localize the deformation in both time and space [Christensen,
1983, 1984]). The along-arc variation in surface heat flow is insignificant (Figure A1b). The magnitude of the
flow in the mantle wedge increases as a whole (Figure A1c). Predicted blueschist-out boundaries are almost
the same as those in the original Case 3 (Figure A1d).
Figure A2 shows the results of Case 3 where the maximum slab-mantle decoupling depth is extended to
120 km from 95 km. We can see that the slab surface temperature at y 300 km becomes larger than that
in the original Case 3 (Figure A2a). It is because the material passes through much deeper and hence
warmer part due to the assumed larger maximum slab-mantle decoupling depth. Surface heat flow does
not show a large along-arc variation neither in this case (Figure A2b). The velocity field shows almost the
same as the original Case 3 except for the region near y 300 km (Figure A2c). Predicted blueschist-out
boundary at y 5 300 km shows a large degree of deepening but still diverges from slab surface with depth,
which cannot fully explain the seismicity below southern Hokkaido (Figure A2d). We also note that the
depth where the blueschist-out boundary starts (108 km) is shallower than the assumed slab-mantle
decoupling depth (120 km) in this cross section.
Acknowledgments
This research was supported by the
Grant-in-Aid for JSPS Research Fellows
(24–7796) to M.M. and by NSF award
EAR-1249353 to P.v.K. We appreciate
discussion with Saeko Kita on
seismological estimate of slab surface
around the junction between Japan
and Kurile arcs. The Generic Mapping
Tools [Wessel and Smith, 1998] were
used to draw the figures in this article.
We thank the reviewers and associate
editor for their detailed comments on
an earlier version of the manuscript.
MORISHIGE AND VAN KEKEN
References
Bengtson, A. K., and P. E. van Keken (2012), Three-dimensional thermal structure of subduction zones: Effects of obliquity and curvature,
Solid Earth, 3, 365–373, doi:10.5194/se-3-365-2012.
Bostock, M., R. Hyndman, S. Rondenay, and S. Peacock (2002), An inverted continental Moho and serpentinization of the forearc mantle,
Nature, 417, 536–538.
Christensen, U. (1983), Convection in a variable-viscosity fluid: Newtonian versus power-law rheology, Earth Planet. Sci. Lett., 64, 153–162.
Christensen, U. (1984), Convection with pressure- and temperature-dependent non-Newtonian rheology, Geophys. J. R. Astron. Soc., 77,
343–384.
Conder, J. (2005), A case for hot slab surface temperatures in numerical viscous flow models in subduction zones with an improved fault
zone parameterization, Phys. Earth Planet. Inter., 149, 155–164.
Cooper, L., D. Ruscitto, T. Plank, P. Wallace, E. Syracuse, and C. Manning (2012), Global variations in H2O/Ce: 1. Slab surface temperatures
beneath volcanic arcs, Geochem. Geophys. Geosyst., 13, Q03024, doi:10.1029/2011GC003902.
Cuvelier, C., A. Segal, and A. A. van Steenhoven (1986), Finite Element Methods and Navier-Stokes Equations, D. Reidel, Dordrecht.
Eberhart-Phillips, D., M. Reyners, M. Chadwick, and G. Stuart (2008), Three-dimensional attenuation structure of the Hikurangi subduction
zone in the central North Island, New Zealand, Geophys. J. Int., 174, 418–434.
Fukao, Y., S. Widiyantoro, and M. Obayashi (2001), Stagnant slab in the upper and lower mantle transition region, Rev. Geophys., 39, 291–
323.
Hacker, B. R. (2008), H2O subduction beyond arcs, Geochem. Geophys. Geosyst., 9, Q03001, doi:10.1029/2007GC001707.
Hall, P. S. (2012), On the thermal evolution of the mantle wedge at subduction zones, Phys. Earth Planet. Inter., 198–199, 9–27, doi:10.1016/
j.pepi.2012.03.004.
Hasegawa, A., J. Nakajima, S. Kita, T. Okada, T. Matsuzawa, and S. Kirby (2007), Anomalous deepening of a belt of intraslab earthquakes in
the Pacific slab crust under Kanto, central Japan: Possible anomalous thermal shielding, dehydration reactions, and seismicity caused
by shallower cold slab material, Geophys. Res. Lett., 34, L09305, doi:10.1029/2007GL029616.
C 2014. American Geophysical Union. All Rights Reserved.
V
2239
Geochemistry, Geophysics, Geosystems
10.1002/2014GC005394
Hayes, G. P., D. J. Wald, and R. L. Johnson (2012), Slab1.0: A three-dimensional model of global subduction zone geometries, J. Geophys.
Res., 117, B01302, doi:10.1029/2011JB008524.
Honda, S. (1985), Thermal structure beneath Tohoku, Northeast Japan: A case study for understanding the detailed thermal structure of
the subduction zone, Tectonophysics, 112, 69–102.
Honda, S., and T. Yoshida (2005), Effects of oblique subduction on the 3-D pattern of small-scale convection within the mantle wedge, Geophys. Res. Lett., 32, L13307, doi:10.1029/2005GL023106.
Hyndman, R., and S. Peacock (2003), Serpentinization of the forearc mantle, Earth Planet. Sci. Lett., 212, 417–432.
Jacques, G., K. Hoernle, J. Gill, F. Hauff, H. Wehrmann, D. Garbe-Schonberg, P. van den Bogaard, I. Bindeman, and L. Lara (2013), Across-arc
geochemical variations in the Southern Volcanic Zone, Chile (34.5–38.0 degrees S): Constraints on mantle wedge and slab compositions, Geochem. Cosmochim. Acta, 123, 218–243.
Karato, S., and P. Wu (1993), Rheology of the upper mantle: A synthesis, Science, 260, 771–778.
Katsumata, A. (2010), Depth of the Moho discontinuity beneath the Japanese islands estimated by traveltime analysis, J. Geophys. Res., 115,
B04303, doi:10.1029/2008JB005864.
Kelemen, P., J. Rilling, E. Parmentier, L. Mehl, and B. Hacker (2003), Thermal structure due to solid-state flow in the mantle wedge beneath
arcs, in Inside the Subduction Factory, Geophys. Monogr., vol. 138, edited by J. Eiler, pp. 293–311, AGU, Washington, D. C.
Kimura, G. (1986), Oblique subduction and collision: Forearc tectonics of the Kuril arc, Geology, 14, 404–407.
Kirby, S. H., W. B. Durham, and L. A. Stern (1991), Mantle phase changes and deep-earthquake faulting in subducting lithosphere, Science,
252, 216–225.
Kita, S., T. Okada, A. Hasegawa, J. Nakajima, and T. Matsuzawa (2010), Anomalous deepening of a seismic belt in the upper-plane of the
double seismic zone in the Pacific slab beneath the Hokkaido corner: Possible evidence for thermal shielding caused by subducted
forearc crust materials, Earth Planet. Sci. Lett., 290, 415–426, doi:10.1016/j.epsl.2009.12.038.
Kita, S., A. Hasegawa, J. Nakajima, T. Okada, T. Matsuzawa, and K. Katsumata (2012), High-resolution seismic velocity structure beneath the
Hokkaido corner, northern Japan: Arc-arc collision and origins of the 1970 m 6.7 Hidaka and 1982 m 7.1 Urakawa-oki earthquakes, J.
Geophys. Res., 117, B12301, doi:10.1029/2012JB009356.
Kneller, E. A., and P. E. van Keken (2008), Effect of three-dimensional slab geometry on deformation in the mantle wedge: Implications for
shear wave anisotropy, Geochem. Geophys. Geosyst., 9, Q01003, doi:10.1029/2007GC001677.
Kneller, E. A., P. E. van Keken, I. Katayama, and S. Karato (2007), Stress, strain, and B-type olivine fabric in the fore-arc mantle: Sensitivity
tests using high-resolution steady-state subduction zone models, J. Geophys. Res., 112, B04406, doi:10.1029/2006JB004544.
Miller, M. S., and B. L. N. Kennett (2006), Evolution of mantle structure beneath the northwest Pacific: Evidence from seismic tomography
and paleogeographic reconstructions, Tectonophysics, 25, TC4002, doi:10.1029/2005TC001909.
Morishige, M., and S. Honda (2013), Mantle flow and deformation of subducting slab at a plate junction, Earth Planet. Sci. Lett., 365, 132–
142, doi:10.1016/j.epsl.2013.01.033.
Morishige, M., S. Honda., and P. J. Tackley (2010), Construction of semi-dynamic model of subduction zone with given plate kinematics in
3D sphere, Earth Planets Space, 62, 665–673, doi:10.5047/eps.2010.09.002.
Nakajima, J., and A. Hasegawa (2006), Anomalous low-velocity zone and linear alignment of seismicity along it in the subducted Pacific
slab beneath Kanto, Japan: Reactivation of subducted fracture zone?, Geophys. Res. Lett., 33, L16309, doi:10.1029/2006GL026773.
Nakajima, J., J. Shimizu, S. Hori, and A. Hasegawa (2006), Shear-wave splitting beneath the southwestern Kurile arc and northeastern Japan
arc: A new insight into mantle return flow, Geophys. Res. Lett., 33, L05305, doi:10.1029/2005GL025053.
Peacock, S., and K. Wang (1999), Seismic consequences of warm versus cool subduction metamorphism: Examples from southwest and
northeast Japan, Science, 286, 937–939.
Plank, T., L. B. Cooper, and C. E. Manning (2009), Emerging geothermometers for estimating slab surface temperatures, Nat. Geosci., 2, 611–
615, doi:10.1038/ngeo614.
Rondenay, S., G. Abers, and P. van Keken (2008), Seismic imaging of subduction zone metamorphism, Geology, 36, 275–278.
Schurr, B., G. Asch, A. Rietbrock, R. Trumbull, and C. Haberland (2003), Complex patterns of fluid and melt transport in the central Andean
subduction zone revealed by attenuation tomography, Earth Planet. Sci. Lett., 125, 105–119.
Stein, C. A., and S. Stein (1992), A model for the global variation in oceanic depth and heat flow with lithospheric age, Nature, 359, 123–
129.
Syracuse, E., P. van Keken, and G. Abers (2010), The global range of subduction zone thermal models, Phys. Earth Planet. Inter., 83, 73–90.
Tanaka, A., M. Yamano, Y. Yano, and M. Sasada (2004), Geothermal gradient and heat flow data in and around Japan (i): Appraisal of heat
flow from geothermal gradient data, Earth Planets Space, 56, 1191–1194.
Tsumura, N., S. Matsumoto, S. Horiuchi, and A. Hasegawa (2000), Three-dimensional attenuation structure beneath the northeastern Japan
arc estimated from spectra of small earthquakes, Tectonophysics, 319, 241–260.
Turner, S., J. Caulfield, M. Turner, P. van Keken, R. Maury, M. Sandiford, and G. Prouteau (2011), Recent contribution of sediments and fluids
to the mantle’s volatile budget, Nat. Geosci., 5, 50–54.
van Keken, P. E., B. Kiefer, and S. M. Peacock (2002), High-resolution models of subduction zones: Implications for mineral dehydration
reactions and the transport of water into the deep mantle, Geochem. Geophys. Geosyst., 3(10), 1056, doi:10.1029/2001GC000256.
van Keken, P. E., et al. (2008), A community benchmark for subduction zone modeling, Phys. Earth Planet. Inter., 171, 187–197.
van Keken, P. E., S. Kita, and J. Nakajima (2012), Thermal structure and intermediate-depth seismicity in the Tohoku-Hokkaido subduction
zones, Solid Earth, 3, 355–364, doi:10.5194/se-3–355-2012.
Wada, I., and K. Wang (2009), Common depth of slab-mantle decoupling: Reconciling diversity and uniformity of subduction zones, Geochem. Geophys. Geosyst., 10, Q10009, doi:10.1029/2009GC002570.
Wada, I., K. Wang, J. He, and R. D. Hyndman (2008), Weakening of the subduction interface and its effects on surface heat flow, slab dehydration, and mantle wedge serpentinization, J. Geophys. Res., 113, B04402, doi:10.1029/2007JB005190.
Wada, I., C. Rychert, and K. Wang (2011), Sharp thermal transition in the forearc mantle wedge as a consequence of nonlinear mantle
wedge flow, Geophys. Res. Lett., 38, L13308, doi:10.1029/2011GL047705.
Watt, S., D. Pyle, T. Mather, and J. Naranjo (2013), Arc magma compositions controlled by linked thermal and chemical gradients above the
subducting slab, Geophys. Res. Lett., 40, 2550–2556, doi:10.1002/grl.50513.
Wessel, P., and W. H. F. Smith (1998), New improved version of the generic mapping tools released, EOS Trans. AGU, 79, 579.
Zhao, D., T. Matsuzawa, and A. Hasegawa (1997), Morphology of the subducting slab boundary in the northeastern Japan arc, Phys. Earth
Planet. Inter., 102, 89–104.
MORISHIGE AND VAN KEKEN
C 2014. American Geophysical Union. All Rights Reserved.
V
2240